Imagine getting a routine chest X-ray and finding out you could be at risk for diabetes. That’s the exciting promise of a new model developed using artificial intelligence (AI).
Scientists say this AI model can look at your X-ray images, even those taken during a routine medical check-up, and warn about possible diabetes risk.
Even if you don’t fit the usual risk factors, this AI model might catch the disease early and help prevent health complications.
A Smart Model for Early Warning
So, how does it work? The researchers taught the AI model to recognize the signs of diabetes risk in X-ray images and electronic health records.
They then tested the model with past health data and found that it could identify patients who were later diagnosed with diabetes. This could mean years of early warning before an actual diagnosis.
It’s worth noting that diabetes cases in the U.S. have more than doubled over the past 35 years.
Currently, medical guidelines suggest diabetes screening for people aged 35 to 70 years old and have a body mass index (BMI) that falls within the overweight to obese range.
However, many people who don’t meet these criteria still develop diabetes. This includes racial and ethnic minorities, where BMI isn’t always a good predictor of diabetes risk.
Left undiagnosed, diabetes can lead to severe health complications, organ damage, and even death.
X-rays: The Opportunity in Every Image
Every year, millions of Americans get chest X-rays for various reasons, such as chest pain, breathing difficulties, injuries, or before surgeries.
These X-ray images become part of a patient’s health record. They might not show diabetes, but the AI model can analyze them later for warning signs of the disease.
To train the AI model, the researchers used over 270,000 X-ray images from 160,000 patients. They used “explainable AI” techniques to understand how the AI model makes its predictions.
They found that the model looked at the location of fatty tissue in the body, which recent medical research has linked with type 2 diabetes, insulin resistance, high blood pressure, and other health problems.
A Promising Result: New Hope for Diabetes Screening
To validate their model, the researchers then tested it on a separate group of about 10,000 patients. They found that the AI model predicted diabetes risk better than models based only on non-image clinical data.
In some cases, the chest X-ray warned of high diabetes risk as early as three years before the patient was diagnosed with the disease.
The AI model also gives a numerical risk score that doctors could use to customize treatment for patients.
“Using chest X-rays to identify those at the highest risk of diabetes, even before a change in blood sugar levels occurs, is a promising method,” says Francisco Pasquel, MD, associate professor in Emory’s Division of Endocrinology, Metabolism and Lipids.
What’s Next: AI in the Health System and Beyond
The researchers’ next step is to further validate the AI model and integrate it into electronic health record systems.
They hope that the model could then alert doctors to pursue traditional diabetes screening for patients flagged as high risk based on X-ray results.
After this, the researchers will explore whether chest X-rays can help diagnose other health conditions, such as vascular disease, heart failure, and chronic obstructive pulmonary disease.
With the help of AI, even a routine chest X-ray could unlock invaluable health insights.
This breakthrough research offers hope for the early detection of diabetes, improving healthcare outcomes, and paving the way for the potential use of AI in detecting other diseases.
If you care about diabetes, please read studies about the sleep problem linked to vision loss in people with diabetes, and this drug may prevent kidney failure in people with diabetes.
For more information about health, please see recent studies that blueberries strongly benefit people with metabolic syndrome, and results showing Vitamin D may reduce dangerous complications in type 2 diabetes.
The study was published in Nature Communications.
Follow us on Twitter for more articles about this topic.
Copyright © 2023 Knowridge Science Report. All rights reserved.
